Analyze videos and images via the Qwen Omni multimodal API.
npx skills add https://github.com/davepoon/buildwithclaude --skill qwen-vision
Please help me install this Agent Skill. Skill: qwen-vision Source: https://github.com/davepoon/buildwithclaude/tree/main/plugins/give-claude-eyes/skills/qwen-vision Command: npx skills add https://github.com/davepoon/buildwithclaude --skill qwen-vision
Claude cannot natively understand video content, so tasks like describing motion, classifying clips, or reviewing footage are impossible without an external multimodal model. This Skill bridges that gap by sending video and image files to the Qwen Omni API and returning text analysis.
Ask Claude to analyze the video at a given file path and describe what happens in it using the qwen-vision skill.
High-intent search queries and answers about installing and using this skill.
Run the qwen_bridge.py script with the video path and a prompt, for example: python3 qwen_bridge.py video.mp4 "Describe what happens". The script sends the file to the Qwen Omni API and prints the analysis as text.
Supported video formats are mp4, mov, avi, mkv, webm, flv, and wmv. Supported image formats are png, jpg, jpeg, gif, webp, bmp, and tiff. Other extensions are rejected with an error.
Get a key from the DashScope console at dashscope.console.aliyun.com or modelstudio.console.alibabacloud.com, then export it as DASHSCOPE_API_KEY. You also need Python 3.9+ with the dashscope package installed via pip.
Lower the --fps flag to sample fewer frames: use --fps 1 for long videos and --fps 0.5 for very long ones. Lower frame rates reduce token usage at the cost of fine motion detail.
Yes. Use --save-context to write the conversation to a JSON file after the first call, then pass --context with that file on subsequent calls. Qwen retains the prior exchange for multi-turn analysis.
Common causes are a missing or invalid DASHSCOPE_API_KEY, exceeded quota, or an unsupported file format. The script prints the API error code and message; verify your key, quota, and file extension first.
joycekushner598-commitsGenerate AI images and videos via a locally deployed Jimeng API.
D3HunterFormat code review findings into severity-sorted JSON with rubric metadata.
darelabs-techStandardizes LLM integration with unified provider interface, caching, rate limiting, and output validation.
GeneralReasoningAnalyze images and sequences with OpenAI GPT vision models into structured JSON.
